Metric for file cache size may become negative

[evenyag]

What type of bug is this?

Incorrect result, User Experience

What subsystems are affected?

Standalone mode, Distributed Cluster

Minimal reproduce step

Run the cluster with the experimental write cache enabled

What did you expect to see?

The metric value should be greater than 0.

What did you see instead?

The value of cached bytes is negative:

What operating system did you use?

Linux

What version of GreptimeDB did you use?

0.9

Relevant log output and stack trace

No response

[evenyag]

Maybe caused by overflow